Abstract
We use a recently found method to characterise all the invertible fourth-order difference equations linear in the extremal values based on the existence of a discrete Lagrangian. We also give some result on the integrability properties of the obtained family and we put it in relation with known classifications. Finally, we discuss the continuum limits of the integrable cases.
Similar content being viewed by others
Notes
In this section and in the next ones, we will indicate various placeholder variables with Greek letters \(\xi \), \(\eta \), \(\zeta \).... We will use these placeholders variables when making statements on functions which might have different arguments, e.g. the function \(g=g\left( \xi \right) \) in Eq. (2.5).
References
M. Bellon, C.-M. Viallet, Algebraic entropy. Comm. Math. Phys. 204, 425–437 (1999)
M. Bruschi, O. Ragnisco, P.M. Santini, G.-Z. Tu, Integrable symplectic maps. Phys. D 49(3), 273–294 (1991)
C.J. Budd, A. Iserles, Geometric integration: numerical solution of differential equations on manifolds. R. Soc. Lond. Philos. Trans. Ser. 357(1754), 945–956 (1999)
C.J. Budd, M.D. Piggott, Geometric integration and its applications, in Handbook of Numerical Analysis, Vol. XI., ed. by F. Cucker (Amsterdam, North-Holland, 2003), pp. 35–139
P. Caldirola, Forze non conservative nella Meccanica Quantistica. Il Nuovo Cimento 18(9), 393–400 (1940)
H.W. Capel, R. Sahadevan, A new family of four-dimensional symplectic and integrable mappings. Phys. A 289, 80–106 (2001)
C. Cresswell, N. Joshi, The discrete first, second and thirty-fourth Painlevé hierarchies. J. Phys. A: Math. Gen. 32, 655–669 (1999)
C. Cresswell, N. Joshi, The discrete Painlevé I hierarchy, in Symmetries and Integrability of Difference Equations, London Mathematical Society Lecture Note Series, ed. by P.A. Clarkson, F.W. Nijhoff (Cambridge University Press, Cambridge, 1999), pp. 197–205
B.A. Dubrovin, A.T. Fomenko, F.T. Novikov, Modern Geometry - Methods and Applications: Part III. Introduction to Homology Theory, 1st edn. (Springer-Verlag, New York, 1990)
S. Elaydi, An Introduction to Difference Equations, 3rd edn. (Springer, Berlin, 2005)
G. Falqui, C.-M. Viallet, Singularity, complexity, and quasi-integrability of rational mappings. Commun. Math. Phys. 154, 111–125 (1993)
M.E. Fels, The inverse problem of the calculus of variations for scalar fourth-order ordinary differential equations. Trans. Am. Math. Soc. 348, 5007–5029 (1996)
H. Flaschka, A.C. Newell, Monodromy- and spectrum-preserving deformations I. Commun. Math. Phys. 76, 65–116 (1980)
R.N. Garifullin, E.V. Gudkova, I.T. Habibullin, Method for searching higher symmetries for quad-graph equations. J. Phys. A: Math. Theor. 44, 325202 (16pp) (2011)
R.N. Garifullin, R.I. Yamilov, Generalized symmetry classification of discrete equations of a class depending on twelve parameters. J. Phys. A: Math. Theor. 45, 345205 (23pp) (2012)
R.N. Garifullin, R.I. Yamilov, Integrable discrete nonautonomous quad-equations as Bäcklund auto-transformations for known Volterra and Toda type semidiscrete equations. J. Phys.: Conf. Ser. 621, 012005 (18pp) (2015)
B. Grammaticos, R.G. Halburd, A. Ramani, C.-M. Viallet, How to detect the integrability of discrete systems. J. Phys. A: Math. Theor. 42, 45400 2 (41 pp) (2009)
G. Gubbiotti, Symmetries and Integrability of Difference Equations: Lecture Notes of the Abecederian School of SIDE 12, Montreal 2016, in Integrability of Difference Equations through Algebraic Entropy and Generalized Symmetries: CRM Series in Mathematical Physics, vol. 3, ed. by D. Levi, R. Verge-Rebelo, P. Winternitz (Springer, Berlin, 2017), pp. 75–152
G. Gubbiotti, On the inverse problem of the discrete calculus of variations. J. Phys. A: Math. Theor. 52, 305203 (29pp) (2019)
Gubbiotti, G., A novel integrable fourth-order difference equation admitting three invariants, Accepted for publication in “Proceedings of the Quantum Theory and Symmetries 11” conference published in CRM Series on Mathematical Physics, (Springer, Berlin, 2020)
Gubbiotti, G., Joshi, N., Tran, D. T., Viallet, C.-M., Complexity and integrability in 4D bi-rational maps with two invariants, Accepted for publication in Springer’s PROMS series: “Asymptotic, Algebraic and Geometric Aspects of Integrable Systems”, arXiv:1808.04942, [nlin.SI] (2018)
G. Gubbiotti, N. Joshi, D.T. Tran, C.-M. Viallet, Bi-rational maps in four dimensions with two invariants. J. Phys. A: Math. Theor. 53, 115201 (24pp) (2020)
G. Gubbiotti, R.I. Yamilov, Darboux integrability of trapezoidal \(H^4\) and \(H^6\) families of lattice equations I: First integrals. J. Phys. A: Math. Theor. 50, 345205 (26pp) (2017)
A. Hagar, Discrete or Continuous?: The Quest for Fundamental Length in Modern Physics (Cambridge University Press, Cambridge, 2014)
J. Hietarinta, N. Joshi, F. Nijhoff, Discrete Systems and Integrability, Cambridge Texts in Applied Mathematics (Cambridge University Press, Cambridge, 2016)
J. Hietarinta, C.-M. Viallet, Singularity confinement and chaos in discrete systems. Phys. Rev. Lett. 81(2), 325–328 (1998)
Hone, A.N.W., Quispel, G.R.W., Analogues of Kahan’s method for higher order equations of higher degree, arXiv:1911.03161 [math.NA] (2019)
P.E. Hydon, E.L. Mansfield, A variational complex for difference equations. Found. Comp. Math. 4, 187–217 (2004)
N. Joshi, C.-M. Viallet, Rational maps with invariant surfaces. J. Int. Syst. 3, xyy017 (14pp) (2018)
E. Kanai, On the quantization of dissipative systems. Proc. Theor. Phys. 3(4), 440–442 (1942)
S.G. Krantz, H.R. Parks, Geometric Integration Theory Cornerstones (Birkhäuser, Boston, 2008)
N.A. Kudryashov, The first and second Painlevé equations of higher order and some relations between them. Phys. Lett. A 224, 353–360 (1997)
D. Levi, R.I. Yamilov, Generalized symmetry integrability test for discrete equations on the square lattice. J. Phys. A: Math. Theor. 44, 145207 (22pp) (2011)
J.D. Logan, First integrals in the discrete variational calculus. Aeq. Math. 9, 210–220 (1973)
S. Maeda, Completely integrable symplectic mapping. Proc. Japan Acad. Ser. A, Math. Sci. 63, 198–200 (1987)
E.M. McMillan, A problem in the stability of periodic systems, in Topics in Modern Physics, A tribute to E.U. Condon, ed. by E. Britton, H. Odabasi (Colorado Assoc. Univ. Press., Boulder, 1971), pp. 219–244
W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannery, Numerical Recipes. Third. The Art of Scientific Computing (Cambridge University Press, Cambridge, 2007), p. xxii+1235
G.R.W. Quispel, J.A.G. Roberts, C.J. Thompson, Integrable mappings and soliton equations. Phys. Lett. A 126, 419 (1988)
G.R.W. Quispel, J.A.G. Roberts, C.J. Thompson, Integrable mappings and soliton equations II. Phys. D 34(1), 183–192 (1989)
D.T. Tran, P.H. van der Kamp, G.R.W. Quispel, Poisson brackets of mappings obtained as \((q,- p)\) reductions of lattice equations. Reg. Chaotic Dyn. 21(6), 682–696 (2016)
A.P. Veselov, Integrable maps. Russ. Math. Surv. 46, 1–51 (1991)
A.P. Veselov, Growth and integrability in the dynamics of mappings. Commun. Math. Phys. 145, 181–193 (1992)
Acknowledgements
GG would like to thank Prof. N. Joshi for the helpful discussions during the preparation of this paper. GG is supported by the Australian Research Council through grants FL120100094 (Prof. N. Joshi), DP190101838 (A/Prof. M. Radnović), and DP200100210 (Prof. N. Joshi and A/Prof. M. Radnović).
Author information
Authors and Affiliations
Corresponding author
Appendices
Proof of Theorem 2
Applying the method described in Sect. 2 to Eq. (2.9a), we obtain following equation (since f and h depend on the same variables we drop the explicit dependence on \(x_{n+1}\), \(x_{n}\) and \(x_{n-1}\)):
Using the CAS Maple 2016 to solve Eq. (A.1), we find that the solution is actually independent of h and has the following form:
Going back to Eq. (2.9a), if we solve with respect to \(\partial ^{3} L_{n-2} /\partial x_{n-2}^{2}\partial x_{n}\) and differentiating with respect to \(x_{n+1}\), we find the following simple compatibility condition:
where we defined:
Equation (A.3) has three factors which can be annihilated separately. The first factor gives \(G\left( x_{n} \right) =0\), that is \(f\equiv 0\), which is not allowed. Therefore, from (A.3) we can choose to fix either f or \(l_{n-2}\). We will now address these two possibilities.
1.1 Fix \(l_{n-2}\) from (A.3)
Solving the second factor in (A.3), we obtain the following value for \(l_{n-2}\):
Inserting (A.5) into (2.9a), we obtain the following equation:
Again we have two factors we can choose to annihilate. The first factors, since no function depends on \(x_{n-2}\) is equivalent to the following equations:
The first equation implies \(G_{-}\left( x_{n-1} \right) =\text {constat}\) and the second one implies \(h=h\left( x_{n+1},x_{n} \right) \). This is not allowed as the equation will be independent of \(x_{n-1}\). Therefore, we are forced to annihilate the second factor. This implies:
Inserting this into (A.5) and using the arbitrariness of \(l_{2,n-2}\), we can write:
Using the definition of \(l_{n-2}\) (A.4) and the fact that discrete Lagrangians are defined only up to total difference, from formula (A.9) we obtain the following form of the Lagrangian:
The Euler–Lagrange equation corresponding to (A.10), upon substitution of Eq. (2.8) are:
Differentiating Eq. (A.11) with respect to \(x_{n+2}\) twice, we obtain:
Using the usual argument, we obtain that we need to annihilate the second factor, which gives:
where \(C_{1,n}\) and \(C_{2,n}\) are two functions depending on n alone. Substituting back in Eq. (A.11) and applying the differential operator
we obtain:
Since \(C_{1,n-2}\ne 0\), we obtain \(G\left( x_{n} \right) =1/K_{1}\) where \(K_{1}\) is a constant. Inserting this value into (A.11), we obtain:
We can take the coefficient with respect to \(x_{n+2}\), and we obtain:
We can rewrite this equation as:
Since \(K_{1}\) is a constant, upon differentiation with respect to \(x_{n+1}\), there exists a constant \(q\in \mathbb {R}\setminus {\left\{ 0 \right\} }\) such that:
Using conditions (A.19) into (A.16), we obtain:
Differentiating with respect to \(x_{n+1}\) and \(x_{n-1}\), we obtain a PDE for h which can be solved to give:
Since h must not depend explicitly on n, we must impose that the coefficient \(F_{n} =C_{1,n-1}/C_{1,n-2}\) is n independent, that is it is a total difference. Using again Eq. (A.19), we obtain:
This implies \(q>0\), that is \(q=\lambda ^{2}\) for some \(\lambda \in \mathbb {R}\setminus { \left\{ 0 \right\} }\), and then
with \(A\in \mathbb {R}\) a constant. However, due to the arbitrariness of \(\lambda \) we can consider only the solution \(C_{1,n}^{+}\). Indeed, \(\lambda \) can be negative and the cases with \(C_{1,n}^{-}\) just follow from the substitution \(\lambda \rightarrow -\lambda \). Therefore, we drop the superscript \(+\) in (A.23). This reasoning implies that the h in (A.21) assumes the following form:
We can finally insert (A.24) into (A.20) and obtain:
Differentiating with respect to \(x_{n+1}\), we obtain a linear PDE for \(l_{2,n}\left( x_{n+1},x_{n} \right) \). Solving such equation, we obtain the following form for this function:
From the form of the Lagrangian function, using the property of equivalence, we can remove the arbitrary function \(l_{2,2,n}\left( x_{n+1} \right) \) and keep only \(l_{2,1,n}\left( x_{n} \right) \). So, in (A.25) this yields:
Differentiation with respect to \(x_{n-1}\) yields the following equation:
Equation (A.28) stimulates the introduction of a potential function \(V=V\left( x_{n},x_{n-1} \right) \) such that:
Using such potential, we have that Eq. (A.28) is identically satisfied, while (A.27) reduces to \(l_{2,1,n}'\left( x_{n} \right) =0\). This implies \(l_{2,1,n}\left( x_{n} \right) = C_{3,n}\), but this function of n can be removed from the Lagrangian as it is a total difference.
Summing up, we obtained that if an additive fourth-order difference Eq. (1.6) is Lagrangian then it has the following form:
Letting \(g\equiv G_{-}\) Eq. (2.5) follows. The constant \(K_{1}\) appearing in the Lagrangian can be scaled away and we obtain the Lagrangian (2.7).
1.2 Fix h from (A.3)
If we fix h from (A.3), we obtain:
After a long calculation which follows the same strategy outlined in the case when we fix \(l_{n-2}\) from (A.3) we find that this case implies \(G\equiv 0\), and so it is impossible.
This concludes the Proof of Theorem 2.
Proof of Theorem 3
Since (3.4) has to be an invariant it must satisfy the following condition:
After substituting the form of Eq. (2.5), no function depends on \(x_{n-2}\), so we can take the coefficients with respect to it. This yields the following system of functional equations which must be identically satisfied:
To solve the above equations, apply again the procedure described in Sect. 2. Applying this strategy to Eq. (B.2a), we are able to solve Eq. (B.2a) fixing the form of the functions \(P_{i}\) in terms of the function g:
In (B.3), the function \(P=P\left( \xi \right) \) is undetermined, and \(C_{i}\) are constants. Inserting (B.3) in Eq. (B.2b), we apply the same strategy with respect to the function \(V=V\left( x_{n+1},x_{n} \right) \) and then with respect to \(P_{4}\left( x_{n+1},x_{n} \right) \). After some steps, we find the following equation:
This equation implies \(C_{2}=0\), as otherwise the function \(g=g\left( \xi \right) \) will be a trivial constant. Substituting \(C_{2}=0\), we obtain the following PDE for \(V=V\left( x_{n},x_{n-1} \right) \):
giving the form of V in terms of g and P:
By arbitrariness of \(V_{1}\left( x_{n-1} \right) \), we can write \(V_{1}\left( x_{n-1} \right) =W\left( x_{n-1} \right) - V_{3}\left( x_{n-1} \right) \) and remove the total difference \(V_{3}\left( x_{n} \right) -V_{3}\left( x_{n-1} \right) \). That is, we can write \(V\left( x_{n},x_{n-1} \right) \) as:
Going back to Eq. (B.2b) and removing iteratively all the functions depending on \(x_{n+1}\) and \(x_{n-1}\), we finally find the following condition on g:
As g needs to be non-trivial and \(C_{1}\ne 0\) from (B.7) we finally obtain that g has to be second-order polynomial of the form (3.5).
Using the conditions in (B.2b), we find the following expression for the function \(V_{2}\):
The function \(V_{2}\left( x_{n} \right) \) appears to be algebraic in \(x_{n}\). However, substituting back in order to check the compatibility conditions we obtain \(A_{4}=0\). Therefore, no algebraic term is left.
The above computations produce rather cumbersome expression for \(P_{4}\left( x_{n},x_{n-1} \right) \), which we will not present. However, we notice that this final form of \(P_{4}\left( x_{n},x_{n-1} \right) \) yield the following condition for the function \(W=W\left( \eta \right) \):
Since \(C_{1}\ne 0\) we perform the scaling \(C_{5}=A_{7}C_{1}\) and \(C_{6}=A_{8}C_{1}\). This yields the expression (3.7) for \(W\left( \eta \right) \) and shows that
This concludes the Proof of Theorem 3.
Rights and permissions
About this article
Cite this article
Gubbiotti, G. Lagrangians and integrability for additive fourth-order difference equations. Eur. Phys. J. Plus 135, 853 (2020). https://doi.org/10.1140/epjp/s13360-020-00858-y
Received:
Accepted:
Published:
DOI: https://doi.org/10.1140/epjp/s13360-020-00858-y